Information processing system, information processing device, information processing method and program

Abstract

To use an output from an intermediate layer of a trained model.SOLUTION: An information processing system includes at least a first information processing device and a second information processing device. The second information processing device transmits an atomic structure to the first information processing device. The first information processing device receives the atomic structure from the second information processing device, inputs the atomic structure to a first model, acquires first information based on an output from an intermediate layer of the first model, and transmits the first information to the second information processing device.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to an information processing system, an information processing device, an information processing method, and a program. [Background technology]

[0002] NNP (Neural Network Potential) is used as a means of acquiring physical properties such as the energy of atoms, molecules, etc., using a neural network model. This NNP is constructed by a neural network model having one or more layers. This neural network model is ultimately constructed as a network that acquires the energy of each atom, so in the intermediate layers, it is possible to acquire information on some properties, physical properties, or information for acquiring physical properties, related to atoms, molecules, etc.

[0003] NNP may be provided in the form of SaaS (Software as a Service). When provided as SaaS, NNP processing is executed on a server, and the output of the intermediate layer of NNP is generally not provided to a client, but may include various information about the atomic structure input as described above. [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] “Neural network embeddings based similarity search method for atomic systems”, Y. Yang, et.al., May 2022, https: / / pubs.rsc.org / en / content / articlelanding / 2022 / DD / D2DD00055E Summary of the Invention [Problem to be solved by the invention]

[0005] One non-limiting problem that embodiments of the present disclosure attempt to solve is utilizing the output from intermediate layers of an inference model. [Means for solving the problem]

[0006] According to one embodiment, an information processing system includes at least a first information processing device and a second information processing device. The second information processing device transmits an atomic structure to the first information processing device. The first information processing device receives the atomic structure from the second information processing device, inputs the atomic structure to a first model, obtains first information based on an output from an intermediate layer of the first model, and transmits the first information to the second information processing device.

[0007] According to one embodiment, the information processing device includes at least one processor. The at least one processor transmits an atomic structure to another information processing device and acquires first information from the other information processing device. The first information is information based on an output from an intermediate layer of the first model, the information being generated by the other information processing device inputting the atomic structure into a first model.

[0008] Also, according to one embodiment, an information processing method is an information processing method in a system having at least a first information processing device and a second information processing device, in which the second information processing device transmits an atomic structure to the first information processing device, the first information processing device receives the atomic structure, inputs the atomic structure to a first model, obtains first information based on output from an intermediate layer of the first model, and transmits the first information to the second information processing device.

[0009] According to one embodiment, an information processing method includes at least one processor transmitting an atomic structure to another information processing device and acquiring first information from the other information processing device, the first information being information based on an output from an intermediate layer of the first model, the information being generated by the other information processing device inputting the atomic structure into a first model. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram illustrating an information processing system according to an embodiment. [Diagram 2] FIG. 2 is a diagram illustrating a process flow of an information processing apparatus according to an embodiment. [Diagram 3] FIG. 2 is a diagram illustrating a process flow of an information processing apparatus according to an embodiment. [Figure 4] FIG. 2 is a diagram illustrating a process flow of an information processing apparatus according to an embodiment. [Diagram 5] FIG. 2 is a diagram illustrating a process flow of an information processing apparatus according to an embodiment. [Figure 6] FIG. 1 is a diagram illustrating an example of implementation of an information processing device according to an embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Problems to be solved by the embodiments of the present disclosure may be, in addition to the problems described above, but are not limited to, examples of problems that may be problems corresponding to the effects described in the embodiments. In other words, a problem that corresponds to at least one of the effects described in the description of the embodiments of the present disclosure may be a problem to be solved by the present disclosure.

[0012] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The drawings and the description of the embodiment are given by way of example only and are not intended to limit the present invention.

[0013] 1 is a diagram showing a schematic diagram of an information processing system according to an embodiment. The information processing system 1 includes, for example, a first information processing device 10 and a second information processing device 20. In this figure, one first information processing device 10 and one second information processing device 20 are shown, but the present invention is not limited to this.

[0014] The information processing system 1 may include, for example, one first information processing device 10 and multiple second information processing devices 20 connected to the first information processing device 10, or multiple first information processing devices 10 and multiple second information processing devices 20 each connected to an arbitrary first information processing device 10 among the multiple first information processing devices 10. Also, a configuration in which one second information processing device 20 is provided for multiple first information processing devices 10 may be used.

[0015] When a plurality of first information processing devices 10 are provided, the first information processing devices 10 may be connected to each other, and when a plurality of second information processing devices 20 are provided, the second information processing devices 20 may be connected to each other. In addition, the connection between any of the first information processing devices 10 and the second information processing devices 20 may be established, for example, via a line such as the Internet, or may be established directly.

[0016] The first information processing device 10 includes a processing circuit 100, a memory circuit 102, and an input / output interface (hereinafter, referred to as an input / output I / F 104). The first information processing device 10 is a device that executes a predetermined calculation based on data received from the second information processing device 20 and transmits the data to the second information processing device 20, and is, for example, a server. In addition to the above, the first information processing device 10 may include a control circuit that controls each component, a power source that supplies power to each component, and the like, as appropriate.

[0017] The processing circuit 100 is a circuit that performs arithmetic processing on data acquired by the first information processing device 10 via the input / output I / F 104. When the information processing system 1 includes a plurality of first information processing devices 10, the processing devices 100 of the plurality of first information processing devices 10 may cooperate with each other to perform arithmetic operations.

[0018] The memory circuit 102 is a circuit that stores information received via the input / output I / F 104. The processing circuit 100 refers to data stored in the memory circuit 102 as necessary. The processing circuit 100 may also store data during a calculation or data resulting from the calculation in the memory circuit 102 as necessary.

[0019] The input / output I / F 104 is an interface that connects the inside and outside of the first information processing device 10. The input / output I / F 104 may include an interface of any standard.

[0020] The first information processing device 10 can provide, for example, SaaS to the second information processing device 20 that transmits data. The detailed calculations of the processing circuit 100 of the first information processing device 10 will be described later.

[0021] The second information processing device 20 includes a processing circuit 200, a memory circuit 202, and an input / output I / F 204. In order to use the service of the first information processing device 10, the second information processing device 20 transmits data to be processed to the first information processing device 10 and receives the result of the calculation from the first information processing device 10. The second information processing device 20 may be, for example, a client corresponding to the first information processing device 10, and in this case, the information processing system 1 can operate as a server-client system. The second information processing device 20 may be a device different from the first information processing device 10, and may be, for example, a server.

[0022] The processing circuit 200 transmits data to be processed to the first information processing device 10 via the input / output I / F 204. In addition, the processing circuit 200 can execute any processing on the calculation result received from the first information processing device 10 via the input / output I / F 204.

[0023] The memory circuit 202 can store the calculation result by the processing circuit 200 of the second information processing device 20, data to be transmitted to the first information processing device 10, or data received from the first information processing device 10. The processing circuit 200 can refer to the data stored in the memory circuit 202 as necessary.

[0024] The input / output I / F 204 is an interface that connects the inside and outside of the second information processing device 20. The input / output I / F 204 may include an interface of any standard.

[0025] Next, a description will be given of the operation of the information processing system 1. The information processing system 1 executes arithmetic processing in the processing circuit 100 of the first information processing device 10 based on data to be processed transmitted from the second information processing device 20, and transmits the result of the arithmetic processing to the second information processing device 20.

[0026] The first information processing device 10 is, for example, a device that executes a neural network-related operation on the received data and transmits the result. More specifically, the first information processing device 10 can receive data on an atomic structure that is an input of a model (e.g., NNP) and transmit the received physical property value on the atomic structure to the second information processing device 20. The first information processing device 10 can also forward propagate the received data on the atomic structure to an intermediate layer of a trained model in the model (e.g., NNP) and transmit the output from this intermediate layer to the second information processing device 20.

[0027] As one example, the data on the atomic structure may include at least information on the type of one or more atoms and the positions (which may be coordinates) of the atoms. For example, based on the data on the atomic structure, each information processing device may obtain information on a compound, a molecular assembly, etc. described by the atomic structure. As another example, the data on the atomic structure may include information on boundary conditions. As yet another example, the data on the atomic structure may include at least information on the nuclei (protons, neutrons) and electrons of one or more atoms.

[0028] The NNP has a trained model that inputs an atomic structure and outputs a physical property value, for example, energy. As a non-limiting example, this trained model is formed with a GNN (Graph Neural Network). For this reason, it is presumed that the output from the intermediate layer of the NNP includes some kind of calculation result from the atomic structure to a physical property value such as energy.

[0029] For this reason, in the present disclosure, the first information processing device 10 at least calculates the forward propagation results up to the intermediate layer before acquiring the physical property values ​​such as energy of the atomic structure received from the second information processing device 20, and transmits them to the second information processing device 20. The second information processing device 20 can continue to execute processing related to the atomic structure based on the acquired output of the intermediate layer.

[0030] Here, the first information processing device 10 may have a processing circuit 100 with a sufficiently higher computation performance than the processing circuit 200 of the second information processing device 20. The processing circuit 100 is, for example, a circuit with a higher parallel computation performance than the processing circuit 200 and capable of executing high-speed computation. The processing circuit 100 is also a circuit that can realize a predetermined computation, for example, a graph computation, faster than the processing circuit 200. The processing circuit 100 of the first information processing device 10 is formed, for example, so as to be able to refer to one or more models related to NNPs, and performs computations up to the output from the intermediate layer upon request from the second information processing device 20, and outputs the results.

[0031] (First embodiment) FIG. 2 is a diagram showing a schematic flow of processing between information processing devices of the information processing system 1 according to an embodiment.

[0032] In the figure, the trained model is shown as a model (NNP) connected from an input layer to an output layer via multiple intermediate layers, but is not limited to this and may be a model including an input layer, one intermediate layer, and an output layer. Also, the intermediate layer that outputs information to be transmitted to the second information processing device 20 may be the intermediate layer next to the input layer, or the intermediate layer immediately before the output layer.

[0033] As mentioned above, the model may be, for example, a GNN, but may also be other neural network models, in which case the model is trained in an appropriate manner based on the type of neural network model.

[0034] In addition, this model is not limited to a neural network model, and may be another model that outputs an appropriate value for an input. In this case, this model may be a model that can output an intermediate value for the output of a final model, or may be one or more models that belong to a model group that obtains a final output from multiple models, and that are used to provide an intermediate output. The same applies to the following embodiments.

[0035] The second information processing device 20 transmits atomic structure data to be processed to the first information processing device 10 via the input / output I / F 204 (S100). The atomic structure data may be described, for example, in a graph format or in another format.

[0036] The first information processing device 10 acquires atomic structure data from the second information processing device 20, and inputs it to the input layer of the trained model developed in the first information processing device 10 (S102).

[0037] The first information processing device 10 forward propagates the atomic structure data from the input layer to the intermediate layer, and obtains the output (first information) from the intermediate layer (S104).

[0038] The first information processing device 10 transmits the output (first information) from the intermediate layer to the second information processing device 20 via the input / output I / F 100 (S106).

[0039] The second information processing device 20 can execute any process based on the data acquired from the first information processing device 10 (S108).

[0040] As described above, the information processing system 1 according to this embodiment makes it possible to form a client-server system capable of quickly acquiring output from the intermediate layer of the NNP. The second information processing device 20 uses SaaS provided by the first information processing device 10 to execute calculations that have high calculation costs in the processing of the NNP in the first information processing device 10, and can quickly acquire output from the intermediate layer.

[0041] The first information processing device 10 may output data up to a predetermined intermediate layer of a predetermined trained model to the second information processing device 20.

[0042] The first information processing device 10 can also selectively use multiple trained models. In this case, the first information processing device 10 can select a trained model to be used based on data received from the second information processing device 20, and transmit the results of forward propagation up to a specified intermediate layer to the second information processing device 20.

[0043] In any of the above cases, the second information processing device 20 may not have information about the trained model. That is, the second information processing device 20 may not have information for forming a model similar to that of the first information processing device 10. Of course, the second information processing device 20 may share parameters of the model used in the second information processing device 20 and the first information processing device 10, and the second information processing device 20 may also be capable of constructing a trained model.

[0044] When the second information processing device 20 has at least some data related to a trained model, the first information processing device 10 can also select a trained model to be used upon request from the second information processing device 20 .

[0045] The first information processing device 10 can transmit the result of forward propagation of input data up to the requested intermediate layer to the second information processing device 20 based on a request from the second information processing device 20 as to which intermediate layer of the trained model data should be obtained. In this case, it is also possible for the user to specify up to which intermediate layer data should be obtained via the second information processing device 20. It is also possible for the user to specify which model to use from a plurality of trained models.

[0046] The first information processing device 10 may transmit information on the trained model used together with the output from the intermediate layer to the second information processing device 20. This model information allows the second information processing device 20 or the user to obtain information on which trained model was used.

[0047] The first information processing device 10 may include, for example, identification information of the trained model (which may include version information, revision information, etc.), information regarding the accuracy of numerical values, and information regarding the device on which the model operates. In addition, the user may specify such information to the first information processing device 10 at the timing of requesting data.

[0048] Second embodiment FIG. 3 is a diagram showing a schematic flow of processing between information processing devices of the information processing system 1 according to an embodiment.

[0049] In this embodiment, the first information processing device 10 forward propagates the received atomic structure through the trained model to obtain an output from the intermediate layer and an output from the output layer (second information) (S104').

[0050] The first information processing device 10 outputs the output data from the intermediate layer acquired in S104' together with the output data from the output layer (second information), typically, for example, an energy value for the atomic structure (S106').

[0051] The second information processing device 20 can acquire output data from each of the intermediate layer and output layer of the trained model in NNP and execute any subsequent processing (S108).

[0052] For example, the second information processing device 20 can perform a similarity calculation or classification of atomic structures, which will be described later, as an arbitrary process thereafter, using information from both the intermediate layer and the output layer.

[0053] By performing processing in this manner, in addition to the above-described embodiment, the second information processing device 20 can execute processing that further refers to data from the output layer.

[0054] Third embodiment The second information processing device 20 can obtain data on what atomic structure the atomic structure is similar to and has similar properties by calculating the similarity in the intermediate layer space forming the NNP, instead of the atomic coordinates themselves contained in the atomic structure, based on the received output from the intermediate layer. The second information processing device 20 can also use the calculated similarity in the intermediate layer space instead of or in addition to the atomic coordinates contained in the atomic structure.

[0055] The atomic structure transmitted by the second information processing device 20 and processed by the first information processing device 10 includes data indicating a structure such as an atomic set (a set of molecules). Therefore, the second information processing device 20 can obtain the similarity between atomic structures in the intermediate layer by using the output from the intermediate layer of the trained model forming the NNP.

[0056] The second information processing device 20 can obtain the similarity of the molecules as a whole, for example, by performing graph matching on a set of outputs from the intermediate layer. The second information processing device 20 can obtain output values ​​from the intermediate layer for various molecular structures via the first information processing device 10. The second information processing device 20 can obtain what atomic structures have similar structures and properties by solving a combinatorial optimization problem of the various molecular structures obtained in this way.

[0057] The second information processing device 20 can also estimate the similarity of atoms, not molecules or compounds, by using the output of the intermediate layer. Similarly to the above, the second information processing device 20 can estimate the similarity of atoms by acquiring the output of the intermediate layer regarding the data of the atomic structure showing the atoms via the first information processing device 10.

[0058] For example, the second information processing device 20 can transmit atomic structures representing various kinds of atoms to the first information processing device 10 and obtain an output from the intermediate layer. In such an information processing system 1, the second information processing device 20 can present to the user data that an atom has a structure or properties similar to those of a certain atom included in the atomic set.

[0059] The output from the intermediate layer is, like the input atomic structure, information on the type of atoms and information on the bonds between atoms, for example, data showing the type of each atom and the bonds between atoms. Therefore, by using the output from the intermediate layer, the second information processing device 20 can obtain information on the structure and properties based on the type of atoms and the bonding state between atoms.

[0060] The second information processing device 20 can also construct a database based on the above, or register data in an existing database.

[0061] The second information processing device 20 can also present to the user the results of a simulation related to the object represented by the input atomic structure, based on the output from the intermediate layer.

[0062] The second information processing device 20 can also implement clustering in the intermediate layer space.

[0063] In the above description, the second information processing device 20 that acquires information from the first information processing device 10 calculates the similarity, but this is not limited to this. For example, the first information processing device 10 may calculate the similarity based on the output from the intermediate layer, and the second information processing device 20 may receive this similarity. Also, for example, the first information processing device 10 may execute a process related to a database, a process related to a simulation, or a process related to clustering.

[0064] (Fourth embodiment) According to the information processing system 1 of the present disclosure, it is also possible to realize potential fitting. For example, the second information processing device 20 can use the output from the intermediate layer of a trained model (first model) deployed in the first information processing device 10 as the input of another trained model (second model).

[0065] FIG. 4 is a diagram showing a schematic flow of processing between information processing devices of the information processing system 1 according to an embodiment.

[0066] The second information processing device 20 acquires output from the intermediate layer of the first model developed in the first information processing device 10 (S106), and can use this output result as input to the input layer of the second model developed in the second information processing device 20 (S110).

[0067] By acquiring output data from the intermediate layer from the first information processing device 10, the second information processing device 20 can input the output data to the input layer as shown in Fig. 4. Furthermore, the second information processing device 20 can use the output data from the intermediate layer acquired by the first information processing device 10 as learning data for a new training target model, for example.

[0068] Of course, if the performance of the processing circuit 200 in the second information processing device 20 is not sufficient, the intermediate layer output of the first model calculated by the first information processing device 10 can be stored in the memory circuit 202 of the second information processing device 20 or an external memory circuit, and the second information processing device 20 can transmit a request for training the second model using another server. In this case, the other server and the first information processing device 10 are not limited to being different servers, and the same information processing device may be used.

[0069] For example, the second information processing device 20 can obtain a characteristic value desired by the user using a second model that outputs a characteristic value desired by the user by using an output from an intermediate layer of a first model in a general-purpose NNP that is appropriately trained by abundant learning data that obtains energy from atomic structures. The obtained characteristic value may be, as a non-limiting example, a characteristic value such as HOMO (Highest Occupied Molecular Orbital), LUMO (Lowest Unoccupied Molecular Orbital), or polarizability.

[0070] By using such an information processing system 1, it is possible to realize appropriate inference processing, etc., by using, for example, a first model in the first information processing device 10, which is versatile and capable of high-precision, high-speed calculations, and a second model that should be kept secret from third parties.

[0071] The first information processing device 10 can also train the second model for the user and notify the user of parameters related to the second model. In this way, a trained model using the output of the intermediate layer of the first model can be customized and provided as SaaS.

[0072] According to this embodiment, the information processing system 1 can provide a specialized NNP for a customer. The information processing system 1 can also realize fitting to data owned by the customer. The parameters of the second model used for fitting can be held in the second information processing device 20, which is a terminal on the user side, or can be held in the first information processing device 10.

[0073] As another example, the second model may be a model that outputs energy.

[0074] Fifth embodiment In each of the above-described embodiments, the first information processing device 10 transmits the result of forward propagation of the trained model to the second information processing device 20, but is not limited to this. The first information processing device 10 may transmit the result of backpropagation of at least some layers of the trained model to the second information processing device 20.

[0075] FIG. 5 is a diagram showing a schematic flow of processing between information processing devices of the information processing system 1 according to an embodiment.

[0076] After the forward propagation, the first information processing device 10 may perform backpropagation from the output layer or an arbitrary intermediate layer to the intermediate layer or input layer before that (S112) and transmit the backpropagation result to the second information processing device 20 (S114). In this way, the first information processing device 10 may calculate the backpropagation output (third information) in an arbitrary layer or between arbitrary layers together with the output from the intermediate layer and transmit it to the second information processing device 20.

[0077] 5 shows a form in which the backpropagation process is executed in the first information processing device 10, but the present invention is not limited to this. When the second information processing device 20 wants to execute backpropagation of a trained model, for example, the first information processing device 10 may transmit information enabling backpropagation of the intermediate layer output (or the output layer output may be included) to the second information processing device 20 together with the intermediate layer output, etc.

[0078] As a simple example, the first information processing device 10 may transmit parameters of a trained model to the second information processing device 20. In this case, the second information processing device 20 can perform backpropagation processing by forming a trained model from the received parameters. By being able to perform backpropagation processing, it is possible to realize re-learning of a trained model in NNP in the second information processing device 20, and as an example, it is also possible to obtain a force value with respect to an energy value (a value obtained by positionally differentiating energy).

[0079] As another example, it is also possible to train an arbitrary neural network model (second model) formed on the second information processing device 20 side, which uses the output of the intermediate layer of a trained model (first model) in NNP as an input, on the first information processing device 10 side. In this case, the second information processing device 20 side learns the second model based on the output from the intermediate layer of the first model, and transmits the learned parameters to the first information processing device 10, thereby executing potential fitting.

[0080] Without being limited to this, during or after the training of the second model formed on the second information processing device 20 side is completed, the results can be used to fit (customize) the first model.

[0081] The second information processing device 20 can, for example, execute backpropagation processing from the output layer of the second model, and transmit the backpropagation value (fourth information) to the input layer of the second model as the backpropagation value of the intermediate layer of the first model to the first information processing device 10 together with information on the atomic structure. The first information processing device 10 can backpropagate the first model based on the information on the atomic structure, the backpropagation value received from the second information processing device 20, and the information on the first model, and execute any processing, for example, potential fitting based on the result of the second model of the first model.

[0082] As another non-limiting example, upon request from the second information processing device 20, the first information processing device 10 can use the backpropagation values ​​of the second model to backpropagate up to the input layer of the first model and calculate the forces acting on the atoms.

[0083] By using the output in this embodiment, the second information processing device 20 can also obtain information such as the position differential for each atom. For example, by using this result, the second information processing device 20 can optimize reaction path estimation and the like as a cost minimization problem in which the difference between the intermediate layer outputs in two certain structures is the cost. By using the position differential information, it is possible to convert the differential of the intermediate layer value into position information (position differential), and it is possible to output a natural path that interpolates between two structures.

[0084] Sixth embodiment The second information processing device 20 can combine the output of the NNP intermediate layer and other data based on the results obtained from the first information processing device 10 to estimate various physical properties related to the transmitted atomic structure.

[0085] For example, the second information processing device 20 can estimate various physical properties based on values ​​registered in an existing database, experimental values, etc. For example, the second information processing device 20 can estimate various physical properties using calculated values, calculation results in other software, and fingerprints. Also, for example, the second information processing device 20 can estimate various physical properties based on the weight, atomic number, etc. of each atom.

[0086] When multiple NNPs are available, the second information processing device 20 can execute any process using output values ​​from the intermediate layers of different NNPs. The different NNPs may be formed in different first information processing devices 10 or may be formed in the same first information processing device 10.

[0087] As another non-limiting example of this embodiment, the second information processing device 20 (or other information processing device) can train the second model using the output from one or more intermediate layers of NNP and the other data described above. The information processing device can train the second model by any machine learning method, for example, using the output from the intermediate layer of NNP and other data as input data and known physical property values ​​as teacher data.

[0088] Seventh embodiment The second information processing device 20 can also generate visualized graphics based on the output from the intermediate layer in the NNP. The second information processing device 20 can convert the output from the intermediate layer itself, or the result of processing this output (including the output input to another trained model), into two or three dimensions with some index as an axis by appropriate processing, and visualize it.

[0089] The second information processing device 20 can, for example, convert a scalar value (n-dimensional value x number of atoms) for each atom acquired as an intermediate layer output, to plot it in a two-dimensional or three-dimensional space.

[0090] Furthermore, the second information processing device 20 can realize processes such as comparison between atoms, comparison between groups of a predetermined number of atoms, comparison between molecules, or comparison between a plurality of molecular structures.

[0091] In addition, the second information processing device 20 can visualize the distribution of the output from the intermediate layer as a histogram. By visualizing, it becomes possible to output a display or the like in a space that is easy for the user to understand.

[0092] Specifically, the second information processing device 20 can calculate the distance between a plurality of molecular structures (atom sets) from the intermediate layer output and plot the distance in a two-dimensional space. As a non-limiting example, the second information processing device 20 can plot the distance in a two-dimensional space using a t-SNE (t-Distributed Stochastic Neighbor Embedding) method. In addition, the second information processing device 20 can visualize the output result from the intermediate layer using any method related to graph display.

[0093] By visualizing in this way, for example, a user can visually confirm groups with close calculated similarities, the results of clustering any physical property based on atomic structure, and the like.

[0094] The second information processing device 20 is capable of visualizing information such as the clustering of oxides and metals as a result of, for example, Bader charge analysis.

[0095] Eighth embodiment The second information processing device 20 can estimate the potential parameters in the process of S108 in Fig. 2 to Fig. 5. For example, as an example of Fig. 4, the second information processing device 20 can train a second model to obtain the potential parameters.

[0096] In this case, the second model is trained as a model that obtains potential parameters from the output of the intermediate layer of the first model. The potential parameters are a model simpler than NNP, and are parameters based on the so-called classical potential. More specifically, they are parameters that contain information on the length and strength of the springs when the bonds between atoms are represented as springs.

[0097] The second information processing device 20 trains the second model by inputting the output of the intermediate layer of the first model to the input layer of the second model and outputting potential parameters from the output layer. The intermediate layer value of the first model has various information about each atom in the atomic structure. Therefore, by forming a model that inputs information about each atom, the second information processing device 20 can train the second model that acquires potential parameters.

[0098] The second information processing device 20 repeats input and forward propagation of the intermediate layer output of the first model to the input layer of the second model, and back propagation processing from the output layer of the second model, so that the output becomes a potential parameter. As a result, the second information processing device 20 can train the second model that outputs a potential parameter when it receives the intermediate layer output of the first model as an input.

[0099] The second information processing device 20 performs training so that the inferred potential parameters match the correct potential parameters. The correct potential parameters may be fitted using those calculated using NNP, or may use already known values.

[0100] As a specific example, the potential parameters for carbon covalent crystals are significantly different between diamond and graphite due to the manner of covalent bonding. When carbon information is input as the atomic structure, the intermediate layer of the NNP outputs data with the properties of diamond when the atomic structure of diamond is input, and outputs data with the properties of graphite when the atomic structure of graphite is input.

[0101] Such potential parameters can be obtained by using the output from the intermediate layer of the NNP. The second information processing device 20 can train the second model using the potential parameters.

[0102] It is also possible to connect this result to the potential fitting in the above embodiment.

[0103] According to each of the above-mentioned embodiments, as a specific example of applying the above disclosure, the information processing system 1 can realize processes such as calculation of similarity of atomic structures based on values ​​output from the NNP intermediate layer, potential fitting, prediction of physical properties, and visualization of quantities related to atomic structures and physical properties.

[0104] In each of the above-described embodiments, the value of the intermediate layer of the NNP is used, but the output from the intermediate layer of the NNP can be, for example, either the value output for each atom or the value output for a combination of atoms, or both values. The value can be appropriately selected depending on the amount to be obtained using the output result of the intermediate layer.

[0105] The value output for the atomic combination is, for example, a value corresponding to the entire atomic structure. The atomic combination is typically a combination of two types of atoms, but is not limited thereto and may be a combination of three or more types of atoms.

[0106] Furthermore, the first information processing device 10 may output the values ​​of the hidden layer of the NNP to the second information processing device 20 through some kind of conversion, rather than directly outputting the values ​​of the hidden layer.

[0107] The first information processing device 10 can output the values ​​of the hidden layer of the NNP through a function that performs linear or nonlinear transformation, for example. As a simple example, the first information processing device 10 may multiply the values ​​of the hidden layer of the NNP by a constant and output them.

[0108] Furthermore, the first information processing device 10 can execute and output statistical processing on values ​​of the intermediate layer of the NNP, for example. The first information processing device 10 may output the results of, for example, Principle Component Analysis (PCA).

[0109] This linear or nonlinear function processing or statistical processing may be realized by some model, for example, a trained neural network model. That is, the first information processing device 10 may input the output from the intermediate layer of the NNP to the trained model, and transmit the output of the trained model to the second information processing device 20.

[0110] In other words, the processing that is executed by the second information processing device 20 in each of the above-mentioned embodiments can also be executed by the first information processing device 10 and the result transmitted to the second information processing device 20.

[0111] In this way, the first information processing device 10 is not limited to transmitting raw data inferred by the NNP, but can also transmit data that has been subjected to some processing on this raw data to the second information processing device 20 .

[0112] In the above embodiment, the second information processing device 20 is a device different from the first information processing device 10, but the present invention is not limited to this.

[0113] For example, the processing executed by the first information processing device 10 in the above-described embodiment may be provided as an application (software) executable on the second information processing device 20, and by executing this application on the second information processing device 20, the processing executed by the first information processing device 10 in the above-described embodiment may be made executable on the second information processing device 20.

[0114] In this case, the application may be an application that executes black box processing, the contents of which are difficult to trace from the second information processing device 20. Also, the application may provide an API (Application Programming Interface) so that the user can use the processing executed by the first information processing device 10 on the second information processing device 20 as appropriate via the API.

[0115] In another embodiment, the first information processing device 10 may execute the processes executed by the second information processing device 20 in the above embodiment, and the second information processing device 20 may acquire the information. In this case, the first information processing device 10 can be used as a SaaS server, as in the above embodiment.

[0116] The first information processing device 10 may provide a user interface, and the user may use the user interface provided by the first information processing device 10 from the second information processing device 20 to request that the process executed by the second information processing device 20 be executed on the first information processing device 10. Alternatively, the request may be written in any file format to the first information processing device 10 by the user, and in this case, the second information processing device 20 can cause the first information processing device 10 to execute the desired process by transmitting a file describing the user's request to the first information processing device 10.

[0117] The trained model in the above embodiment may be a concept that includes, for example, a model that has been trained as described and then further distilled using a general method.

[0118] A part or all of each device (first information processing device 10 or second information processing device 20) in the above-mentioned embodiment may be configured with hardware, or may be configured with information processing of software (program) executed by a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) or the like. In the case of being configured with information processing of software, software that realizes at least a part of the functions of each device in the above-mentioned embodiment may be stored in a non-transient storage medium (non-transient computer-readable medium) such as a CD-ROM (Compact Disc-Read Only Memory) or a USB (Universal Serial Bus) memory, and the software information processing may be executed by reading the software into a computer. The software may also be downloaded via a communication network. Furthermore, the information processing by the software may be executed by hardware by implementing all or a part of the processing of the software in a circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0119] The storage medium that stores the software may be a removable medium such as an optical disk, or a fixed medium such as a hard disk or memory. The storage medium may be provided inside the computer (such as a main storage device or auxiliary storage device) or outside the computer.

[0120] 6 is a block diagram showing an example of a hardware configuration of each device (the first information processing device 10 or the second information processing device 20) in the above-mentioned embodiment. Each device may be realized as a computer 7 including, for example, a processor 71, a main storage device 72 (memory), an auxiliary storage device 73 (memory), a network interface 74, and a device interface 75, which are connected via a bus 76.

[0121] Although the computer 7 in FIG. 6 includes one of each component, it may include multiple of the same components. Although one computer 7 is shown in FIG. 6, the software may be installed in multiple computers, and each of the multiple computers may execute the same or different parts of the software. In this case, it may be a distributed computing form in which each computer communicates via a network interface 74 or the like to execute the processing. In other words, each device (the first information processing device 10 or the second information processing device 20) in the above-mentioned embodiment may be configured as a system in which one or more computers execute instructions stored in one or more storage devices to realize the function. Also, it may be configured such that information transmitted from a terminal is processed by one or more computers provided on the cloud, and the processing results are transmitted to the terminal.

[0122] Various calculations of each device (first information processing device 10 or second information processing device 20) in the above-mentioned embodiment may be executed in parallel using one or more processors, or using multiple computers via a network. Also, various calculations may be distributed to multiple arithmetic cores in a processor and executed in parallel. Also, a part or all of the processes, means, etc. disclosed herein may be realized by at least one of a processor and a storage device provided on a cloud that can communicate with the computer 7 via a network. Thus, each device in the above-mentioned embodiment may be in the form of parallel computing using one or more computers.

[0123] The processor 71 may be an electronic circuit (processing circuit, processing circuitry, CPU, GPU, FPGA, ASIC, etc.) that at least controls a computer or performs calculations. The processor 71 may be any of a general-purpose processor, a dedicated processing circuit designed to execute a specific calculation, or a semiconductor device including both a general-purpose processor and a dedicated processing circuit. The processor 71 may include an optical circuit, or may include a calculation function based on quantum computing.

[0124] The processor 71 may perform arithmetic processing based on data or software input from each device or the like configured inside the computer 7, and may output arithmetic results or control signals to each device or the like. The processor 71 may control each component constituting the computer 7 by executing the OS (Operating System) of the computer 7, applications, or the like.

[0125] Each device (first information processing device 10 or second information processing device 20) in the above-described embodiment may be realized by one or more processors 71. Here, the processor 71 may refer to one or more electronic circuits arranged on one chip, or may refer to one or more electronic circuits arranged on two or more chips or two or more devices. When multiple electronic circuits are used, the electronic circuits may communicate with each other by wire or wirelessly.

[0126] The main memory device 72 may store instructions executed by the processor 71 and various data, and information stored in the main memory device 72 may be read by the processor 71. The auxiliary memory device 73 is a memory device other than the main memory device 72. Note that these memory devices refer to any electronic components capable of storing electronic information, and may be semiconductor memories. The semiconductor memories may be either volatile or nonvolatile memories. The memory devices for saving various data in each device (the first information processing device 10 or the second information processing device 20) in the above-mentioned embodiments may be realized by the main memory device 72 or the auxiliary memory device 73, or may be realized by an internal memory built into the processor 71. For example, the memory circuit 102 or the memory circuit 202 in the above-mentioned embodiments may be realized by the main memory device 72 or the auxiliary memory device 73.

[0127] When each device (first information processing device 10 or second information processing device 20) in the above-mentioned embodiment is composed of at least one storage device (memory) and at least one processor connected (coupled) to the at least one storage device, at least one processor may be connected to one storage device. At least one storage device may be connected to one processor. Also, a configuration in which at least one processor among a plurality of processors is connected to at least one storage device among a plurality of storage devices may be included. This configuration may also be realized by storage devices and processors included in a plurality of computers. Furthermore, a configuration in which a storage device is integrated with a processor (for example, a cache memory including an L1 cache and an L2 cache) may be included.

[0128] The network interface 74 is an interface for connecting to the communication network 8 wirelessly or by wire. The network interface 74 may be an appropriate interface, such as one conforming to an existing communication standard. The network interface 74 may exchange information with an external device 9A connected via the communication network 8. The communication network 8 may be any one of a WAN (Wide Area Network), a LAN (Local Area Network), a PAN (Personal Area Network), etc., or a combination thereof, as long as information is exchanged between the computer 7 and the external device 9A. An example of a WAN is the Internet, an example of a LAN is IEEE 802.11 or Ethernet (registered trademark), and an example of a PAN is Bluetooth (registered trademark) or NFC (Near Field Communication), etc.

[0129] The device interface 75 is an interface such as a USB that directly connects to an external device 9B.

[0130] The external device 9A is a device connected to the computer 7 via a network. The external device 9B is a device directly connected to the computer 7.

[0131] The external device 9A or the external device 9B may be, for example, an input device. The input device is, for example, a device such as a camera, a microphone, a motion capture device, various sensors, a keyboard, a mouse, or a touch panel, and provides acquired information to the computer 7. Alternatively, the external device 9A or the external device 9B may be a device having an input unit, a memory, and a processor, such as a personal computer, a tablet terminal, or a smartphone.

[0132] Moreover, the external device 9A or the external device 9B may be, for example, an output device. The output device may be, for example, a display device such as an LCD (Liquid Crystal Display) or an organic EL (Electro Luminescence) panel, or a speaker that outputs sound or the like. Alternatively, the output device may be a device including an output unit, a memory, and a processor, such as a personal computer, a tablet terminal, or a smartphone.

[0133] In addition, the external device 9A or the external device 9B may be a storage device (memory). For example, the external device 9A may be a network storage or the like, and the external device 9B may be a storage device such as an HDD.

[0134] Furthermore, the external device 9A or the external device 9B may be a device having some of the functions of the components of each device (the first information processing device 10 or the second information processing device 20) in the above-mentioned embodiment. In other words, the computer 7 may transmit some or all of the processing results to the external device 9A or the external device 9B, or may receive some or all of the processing results from the external device 9A or the external device 9B.

[0135] In this specification (including the claims), when the phrase "at least one of a, b, and c" or "at least one of a, b, or c" (including similar phrases) is used, it includes any of a, b, c, a-b, a-c, b-c, or a-b-c. It may also include multiple instances of any element, such as a-a, a-b-b, a-a-b-b-c-c, etc. It also includes the addition of elements other than the enumerated elements (a, b, and c), such as having d, as in a-b-c-d.

[0136] In this specification (including claims), when expressions such as "using data as input / based on / according to / in response to data" (including similar expressions) are used, unless otherwise specified, this includes the use of data itself or data that has been processed in some way (e.g., data with noise added, normalized, features extracted from data, intermediate representation of data, etc.). In addition, when a statement is made that a result is obtained "using data as input / based on / according to / in response to data" (including similar expressions), unless otherwise specified, this includes the use of the result based only on the data in question or the use of the result influenced by other data, factors, conditions, and / or states other than the data in question. In addition, when a statement is made that "data is output" (including similar expressions), unless otherwise specified, this includes the use of data itself as output or data that has been processed in some way (e.g., data with noise added, normalized, features extracted from data, intermediate representation of data, etc.) as output.

[0137] In this specification (including the claims), the terms "connected" and "coupled" are intended as open-ended terms including any of direct connection / coupling, indirect connection / coupling, electrically connection / coupling, communicatively connection / coupling, functionally connection / coupling, physically connection / coupling, etc. The terms should be interpreted appropriately according to the context in which the terms are used, but any connection / coupling form that is not intentionally or naturally excluded should be interpreted as being included in the terms without any limitation.

[0138] In this specification (including the claims), when the expression "A configured to B" is used, it may include that the physical structure of element A has a configuration capable of performing operation B, and that the permanent or temporary setting / configuration of element A is configured / set to actually perform operation B. For example, when element A is a general-purpose processor, it is sufficient that the processor has a hardware configuration capable of performing operation B, and is configured to actually perform operation B by setting a permanent or temporary program (instruction). Also, when element A is a dedicated processor or dedicated arithmetic circuit, it is sufficient that the circuit structure of the processor is implemented to actually perform operation B, regardless of whether control instructions and data are actually attached.

[0139] When used in this specification (including the claims), terms implying containing or possessing (e.g., "comprising / including" and "having") are intended to be open-ended terms that include the inclusion or possession of things other than the object indicated by the object of the term. When the object of such terms implies no quantity or a singular number (such as an article such as "a" or "an"), the expression should be construed as not being limited to a specific number.

[0140] In this specification (including the claims), even if expressions such as "one or more" or "at least one" are used in some places and expressions that do not specify a quantity or suggest a singular number (expressions using the articles a or an) are used in other places, the latter expressions are not intended to mean "one." In general, expressions that do not specify a quantity or suggest a singular number (expressions using the articles a or an) should be interpreted as not necessarily being limited to a specific number.

[0141] In this specification, when a particular advantage / result is described as being obtained from a particular configuration of an embodiment, it should be understood that the same effect can also be obtained from one or more other embodiments having the same configuration, unless there is a reason to the contrary. However, it should be understood that the presence or absence of the effect generally depends on various factors, conditions, and / or states, and that the effect is not necessarily obtained by the configuration. The effect is merely obtained by the configuration described in the embodiment when various factors, conditions, and / or states are satisfied, and the effect is not necessarily obtained in the invention according to the claim that specifies the configuration or a similar configuration.

[0142] In this specification (including the claims), when a term such as "maximize" is used, it includes finding a global maximum, finding an approximation of a global maximum, finding a local maximum, and finding an approximation of a local maximum, and should be interpreted accordingly according to the context in which the term is used. It also includes finding approximations of these maxima probabilistically or heuristically. Similarly, when a term such as "minimize" is used, it includes finding a global minimum, finding an approximation of a global minimum, finding a local minimum, and finding an approximation of a local minimum, and should be interpreted accordingly according to the context in which the term is used. It also includes finding approximations of these minima probabilistically or heuristically. Similarly, when a term such as "optimize" is used, it includes finding a global optimum, finding an approximation of a global optimum, finding a local optimum, and finding an approximation of a local optimum, and should be interpreted accordingly according to the context in which the term is used. It also includes finding approximations of these optimum values ​​probabilistically or heuristically.

[0143] In this specification (including claims), when multiple pieces of hardware perform a predetermined process, each piece of hardware may cooperate to perform the predetermined process, or a part of the hardware may perform all of the predetermined process. Also, a part of the hardware may perform a part of the predetermined process, and another piece of hardware may perform the rest of the predetermined process. In this specification (including claims), when an expression such as "one or more pieces of hardware perform a first process, and the one or more pieces of hardware perform a second process" (including similar expressions) is used, the hardware performing the first process and the hardware performing the second process may be the same or different. In other words, it is sufficient that the hardware performing the first process and the hardware performing the second process are included in the one or more pieces of hardware. The hardware may include an electronic circuit, or a device including an electronic circuit.

[0144] In this specification (including the claims), when multiple storage devices (memories) store data, each of the multiple storage devices may store only a portion of the data, or may store the entire data. Also, a configuration in which only some of the multiple storage devices store data may be included.

[0145] The above-described embodiment can also be summarized as follows.

[0146] (1) An information processing system including at least a first information processing device and a second information processing device, The second information processing device is transmitting the atomic structure to the first information processing device; The first information processing device is receiving the atomic structure from the second information processing device; inputting said atomic structure into a first model; obtaining first information based on an output from a hidden layer of the first model; Transmitting the first information to the second information processing device; Information processing system.

[0147] The atomic structure is a concept including data related to the atomic structure. The first information may be at least the value of the intermediate layer itself or the value of the intermediate layer converted by a predetermined process.

[0148] (2) The first information processing device is Further obtaining second information based on an output from an output layer of the first model; Transmitting the second information to the second information processing device; An information processing system according to (1).

[0149] The second information may be at least either the value of the output layer itself or the value of the output layer transformed by a predetermined process.

[0150] (3) The first information processing device is performing a backpropagation process on the first model to obtain further third information; transmitting the third information to the second information processing device; An information processing system according to (1) or (2).

[0151] (4) The second information processing device is Obtaining the first information; inputting the first information into a second model to estimate a predetermined physical property value for the atomic structure; An information processing system according to any one of (1) to (3).

[0152] (5) the predetermined physical property value is an energy for the atomic structure; (4) An information processing system according to the above.

[0153] (6) The first model is a model that forms NNP (Neural Network Potential), An information processing system according to any one of (1) to (5).

[0154] (7) at least one processor; The at least one processor: Transmitting the atomic structure to another information processing device; Acquire first information from the other information processing device; The first information is information based on an output from an intermediate layer of the first model, the information being generated by inputting the atomic structure into a first model by the other information processing device. Information processing device.

[0155] The information based on the output from the intermediate layer may be at least either the value of the intermediate layer itself or the value of the intermediate layer transformed by a predetermined process.

[0156] (8) The at least one processor: Obtaining the first information; inputting the first information into a second model to estimate a predetermined physical property value for the atomic structure; (7) An information processing device.

[0157] (9) the predetermined physical property value is an energy for the atomic structure; An information processing device according to (8).

[0158] (10) The at least one processor: acquiring information regarding at least a part of the first model from the other information processing device; performing a backpropagation process using the received information about at least a portion of the first model and an output from an output layer of the second model; An information processing device according to (8) or (9).

[0159] (11) The at least one processor: calculating a similarity between at least a part of the atomic structure and at least a part of another atomic structure based on the received first information; An information processing device according to any one of (7) to (10).

[0160] The similarity calculation may be for the entire atomic structure, or may be for a part of the atomic structure.

[0161] (12) The at least one processor: Estimating potential parameters by inputting the first information into a second model; An information processing device according to any one of (7) to (11).

[0162] (13) The at least one processor: Displaying the first information on a display device. An information processing device according to any one of (7) to (12).

[0163] (14) The first information displayed on the display device is lower-dimensional information than an output from an intermediate layer of the first model. An information processing device according to (13).

[0164] (15) The first model is a model that forms NNP (Neural Network Potential), An information processing device according to any one of (7) to (14).

[0165] (16) An information processing method in a system including at least a first information processing device and a second information processing device, The second information processing device transmitting the atomic structure to the first information processing device; The first information processing device receiving the atomic structure; inputting said atomic structure into a first model; obtaining first information based on an output from a hidden layer of the first model; Transmitting the first information to the second information processing device; Information processing methods.

[0166] The above information processing method can have any one of the features described in (2) to (16).

[0167] (17) At least one processor Transmitting the atomic structure to another information processing device; acquiring first information from the other information processing device; 1. A method comprising: the first information is information based on an output from an intermediate layer of the first model, the information being generated by the other information processing device inputting the atomic structure into the first model; Information processing methods.

[0168] The above information processing method can have any one of the features described in (2) to (16).

[0169] Although the embodiments of the present disclosure have been described in detail above, the present disclosure is not limited to the individual embodiments described above. Various additions, modifications, replacements, and partial deletions are possible within the scope of the conceptual idea and intent of the present disclosure derived from the contents defined in the claims and their equivalents. For example, in the above-mentioned embodiments, when numerical values ​​or formulas are used for explanation, these are shown for illustrative purposes and do not limit the scope of the present disclosure. In addition, the order of each operation shown in the embodiment is also illustrative and does not limit the scope of the present disclosure. [Explanation of symbols]

[0170] 1: Information processing system, 10: First information processing device, 100: processing circuit, 102: Memory circuit, 104: Input / Output I / F, 20: second information processing device, 200: Processing circuit, 202: Memory circuit, 204: Input / Output Interface

Claims

1. An information processing system comprising at least a first information processing device and a second information processing device, The second information processing device is The atomic structure is transmitted to the first information processing device, The first information processing device is The atomic structure is received from the second information processing device, The atomic structure is input into the first model. First information is obtained based on the output from the intermediate layer of the first model. The first information is transmitted to the second information processing device, The second information processing device is The first information received from the first information processing device is input to the second model to obtain values ​​related to the atomic structure. Information processing system.

2. The first information processing device is Further information is obtained based on the output from the output layer of the first model, The second information is transmitted to the second information processing device. The information processing system according to claim 1.

3. The first information processing device is The third information is further obtained by performing the backpropagation process of the first model described above, The third information is transmitted to the second information processing device. The information processing system according to claim 1.

4. The aforementioned values ​​are physical properties related to the atomic structure. The information processing system according to claim 1.

5. The aforementioned physical property is the energy relating to the atomic structure. The information processing system according to claim 4.

6. The second information processing device is Information relating to at least a part of the first model is obtained from the first information processing device. The backpropagation process is performed using the information obtained from at least a portion of the first model and the output from the output layer of the second model. The information processing system according to claim 1.

7. The second information processing device is Based on the received first information, the degree of similarity between at least a portion of the atomic structure and at least a portion of other atomic structures is calculated. The information processing system according to claim 1.

8. The second information processing device causes the first information to be displayed on the display device. The information processing system according to claim 1.

9. The aforementioned value is a potential parameter relating to the atomic structure. An information processing system according to any one of claims 1 to 8.

10. The aforementioned first model is a model that forms an NNP (Neural Network Potential). An information processing system according to any one of claims 1 to 8.

11. The first information includes at least the output from the intermediate layer or a value obtained by converting the output from the intermediate layer. An information processing system according to any one of claims 1 to 8.

12. At least one memory, Equipped with at least one processor, The aforementioned at least one processor is The atomic structure is transmitted to other information processing devices. First information is obtained from the aforementioned other information processing device, By inputting the first information into the second model, values ​​related to the atomic structure are obtained. The first information is information based on the output from the intermediate layer of the first model, which is generated when the other information processing device inputs the atomic structure into the first model. Information processing device.

13. The aforementioned values ​​are physical properties related to the atomic structure. The information processing apparatus according to claim 12.

14. The aforementioned physical property is the energy relating to the atomic structure. The information processing apparatus according to claim 13.

15. The aforementioned at least one processor is Information relating to at least a part of the first model is obtained from the other information processing device, Using the information received about at least a portion of the first model, a backpropagation process is performed using the output from the output layer of the second model. The information processing apparatus according to claim 12.

16. The aforementioned at least one processor is Based on the received first information, the degree of similarity between at least a portion of the atomic structure and at least a portion of other atomic structures is calculated. The information processing apparatus according to claim 12.

17. The aforementioned at least one processor is The above-mentioned first information is displayed on the display device. The information processing apparatus according to claim 12.

18. The first information displayed on the display device is lower-dimensional information than the output from the intermediate layer of the first model. The information processing apparatus according to claim 17.

19. The aforementioned value is a potential parameter relating to the atomic structure. An information processing apparatus according to any one of claims 12 to 18.

20. The previous model 1 is a model that forms an NNP. An information processing apparatus according to any one of claims 12 to 18.

21. The first information includes at least the output from the intermediate layer or a value obtained by converting the output from the intermediate layer. An information processing apparatus according to any one of claims 12 to 18.

22. At least one processor, The atomic structure is transmitted to other information processing devices. First information is obtained from the aforementioned other information processing device, By inputting the aforementioned first information into the second model, values ​​related to the atomic structure are obtained. It is a method, The first information is information based on the output from the intermediate layer of the first model, which is generated when the other information processing device inputs the atomic structure into the first model. Information processing methods.

23. An information processing system according to any one of claims 1 to 8 acquires values ​​relating to atomic structure. Information processing methods.

24. To cause an information processing system comprising at least a first information processing device and a second information processing device to execute the information processing method described in claim 23, program.

25. The information processing device according to any one of claims 12 to 18 acquires a value relating to atomic structure. Information processing methods.

26. To cause at least one processor to execute the information processing method described in claim 25, program.

Images